What makes an unsteady graph unsteady or why oil is oil? - page 31

 
FOXXXi >>:
Нет,не так - тогда бы это был бы идеальный тренд.Например сумма 16-ти пятнадцатиминутных приращений СБ будет давать приращение одного часа.

No, exactly. The sum of 16 15-minute increments for any given series is a 4-hour increment. What is meant is that one random variable is treated as the sum of several others, which are obtained from the time series of the lower timeframe by time shifting. if the mixing operator gives independent variables, the convolution condition must be met, and it de facto is not.

illustration:

 
FOXXXi писал(а) >>
Reread what, your questions to me?

I'm tired of proving anything. Let it be your way.

Reshetov wrote >>.

If p = q, then there is a proven theorem about returning a point to the initial (any historical) value with probability 1. Hence any channel can be drawn, through the levels already present on the history and bluntly following the counter-trend inside the channel. The arc sine theorem should also be taken into account, i.e. prices may take quite a long time to return to the initial levels. And the time intervals between returns will systematically increase.

A profitable strategy to play the game of foxholes?

 
lea >>:

Прибыльная стратегия игры в орлянку?

In this theorem, the initial value can be replaced by any other value, and the proof will not undergo a single change. Therefore, in order to make a game of beaucoup profitable, one must theoretically have an infinite deposit:)
 
alsu >>:
нет, именно так. сумма 16 15-минутных приращений для любого конкретного ряда - это приращение 4-х часов. Имеется ввиду, что одна случайная величина рассматривается как сумма нескольких других, которые получаются из временного ряда младшего таймфрейма смещением по времени. если оператор смешения дает независимые величины, должно выполняться условие свертки, а оно де-факто не выполняется.

For which series? In this case we consider the euro/dollar - SB. It will not be de facto fulfilled, because sqrt(2minutes) = 1.41, not 2 minutes.

In other words, according to you it turns out that 16 15-minute increments are directed in one direction - great!

 
FOXXXi >>:

Да для какого любого ряда?В данном случае рассматриваем евро/долл - СБ.Оно и небудет де-факто выполняться,потому что sqrt(2минут) = 1,41,а не 2 минутам.

Т.е. по твоему получается что 16-ть пятнадцатиминутных приращений направлены в одну сторону - замечательно!

Once again, the algebraic sum takes into account the sign of the increments. If the exchange rate went up by 100 points in the first half-hour and dropped by 50 points in the second half-hour, the resulting one-hour increment is 100-50=50 points. What else is unclear? Don't they still teach negative numbers in school?
 
alsu >>:
еще раз - алгебраическая сумма учитывает знак приращений. Если в первые полчаса курс вырос на 100 пунктов, а во вторые упал на 50 результирующее приращение за час 100-50=50 пунктов. Чего еще непонятно? Отрицательные числа в школе вроде еще изучают?
In my case, I am talking about the amount of increment, which you are evaluating, i.e. the average increment in 4 hours is not 16 times but 4 times greater than the average increment in 15 minutes - this is what I wanted to tell you.
 
gpwr >>:


Я уверен что производил вычисления правильно, коллега.

I didn't doubt that, the question is what exactly you calculated, I'm trying to figure it out (for myself only).

Instead of a bunch of segments, I chose two.

The thing is that clear (to me) and proven methods of verification, for some reason, requires a larger number of segments, simply requires a number. The obtained series of parameters by segments is analysed for correspondence to a certain (depending on the method or its variant) distribution and only after that one can apply the trend criteria. It is difficult to draw such conclusions for two points.

Of course, if you want to, you can. Here is a simple example: EURUSD series, M15, with a history of 200,000 samples. I divide the series into two parts of 100 000 and plot the frequencies of the first differences (the second picture is a logarithm):


I think you will smile, but the visual analysis for stationarity estimation also applies as the first information. Let's see how the RMS of the two chunks correlate:


I don't know how you look at it - but it is quite obvious that the processes are stationary, and the RMS is the same to the sixth digit. In general, it is a stationary process, and statistical methods confirm that with very good accuracy (and it works on smaller scales). Another thing is that this in itself does not make the process predictable.

The number of bars in both segments is about 22 thousand, the data is from October 2006. Of course, I can go deeper into history, but I do not have such a long history. The result is not entirely unexpected. The first moment of the price difference goes around zero, which indicates that there is no trend. The second momentum rises from 1 to 3.4 as we move from the first segment to the end of the second segment. It is also clear. The first segment includes prices for 2006-2008 and the second segment includes prices for 2008-2010.

History is an important thing, and my understanding of history is that it is something that provides knowledge. But quoting is a very "slow" process, if I may say so, and we will never "wait" for all the knowledge of the process, no matter how long we observe it. But that's just a word of reflection.

Market volatility has increased substantially during this time. To assume that the second moment of price difference is constant in time is to assume that there are no periods of economic calm (growth) and crises in the world, which we are going through now. This is on larger timeframes. On shorter timeframes, volatility (or variance) is also not constant due to news that causes spikes and gaps in prices at certain hours of the day and days of the week. So I can understand why there is no stationarity in the market even without mathematics, even if you take the first price difference. Although, I'm willing to change my mind if someone presents their calculations indicating that the price difference is stationary at least in the broad sense.

I don't like that kind of reasoning - it's not about anything. It's empty philosophy and one can roll out the opposite theory, just for the sake of sport. As for trajectory deviations due to heavy tails with all those gaps, jumps and so on, that's a different story.

By the way, I am not claiming that there are no short periods of stationarity in prices. For example EURGBP behaves quite calm during the European night time, without any significant trends and volatility bursts due to the lack of news regarding both currencies at that time. This pair is a heaven for pip traders. Just ask YuraZ :-)

I have given above some arguments that I had enough time to explain. Everything else was eaten up by a virus a few years ago. You see, I'm not going to argue, but I have made my own conclusions and as I see it, they are well-founded with all due precision. But the trick is very simple and it's quite different.

 

lea писал(а) >>

...........................

A profitable strategy for playing beagle?

Mm-hmm. Portfolio strategy. In 30 back alleys at a time. Better yet, a hundred...
 
FOXXXi >>:
В моём случае речь идёт о размере приращения,ты же ведь это оцениваешь.Тоесть ср. приращение за 4 часа будет больше ср. приращения за 15 мин. не в 16 раз,а в 4 раза - это я и хотел тебе сказать.

OK, you obviously haven't figured it out.

I'm not estimating the average increments, or even their size. I'm estimating the probability density function of price increments - there is a corresponding apparatus for that.

I just wanted to say that the method of checking the series for independence of increments given by me gives practically unambiguous and 99.99% theoretically correct result - the price series is not the series with independent increments (even if they don't correlate much or at all). This, in turn, suggests that all price models which imply independence of neighbouring counts are inadequate.

Moreover, statistical dependence between adjacent readings appears to be the same in form on series charts of at least several lower timeframes (I checked it with sufficient accuracy up to H4) (though it must be proved - and there is just not enough historical data for that). I.e., it seems that the specified dependence is not a random phenomenon, at least partially - and therefore it can be predicted.

Once again, this conclusion is theoretical and based solely on mathematics, no speculations from the field of "fundamental analysis":)

 
alsu >>:

ОК, видно не разобрались.

Я оцениваю не средние приращения и даже не их размер. Я оцениваю вид плотности распределения вероятностей приращения цены - для этого имеется соответствующий матаппарат.

Я только хотел сказать, что приведенная мной методика проверки ряда на независимость приращений дает практически однозначный и теоретически на 99,99% обоснованный результат - ценовой ряд не является рядом с независимыми приращениями (даже если они мало или вообще не коррелируют). А это, в свою очередь, говорит о том, что все модели работы с ценой, подразумевающие независимость соседних отсчетов - неадекватны.

Более того, по-видимому (хотя это еще и надо доказать - а для этого просто не хватает исторических данных), статистическая зависимость между соседними отсчетами одинакова по форме на графиках ряда, по крайней мере, младших таймфреймов (вплоть до Н4 я это проверял с достаточной точностью). Т.Е. похоже на то, что указанная зависимость - явление неслучайное, как минимум, отчасти - а значит может быть спрогнозировано.

Еще раз повторюсь - этот вывод теоретический и основан исключительно на математике, никаких домыслов из области "фундаментального анализа":)

A powerful statement, and the main thing is that everyone subconsciously wants it to be true.